A dynamic planning method for ship routing based on port throughput
By serializing and clustering analysis of port throughput data in ship path planning, combined with the timing prediction model, real-time dynamic updates are achieved, which solves the problem that static planning is difficult to avoid port congestion, and improves the efficiency and effectiveness of ship path planning .
Patent Information
- Application Number
- CN202310271370.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-17
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2043-03-17
AI Technical Summary
In the prior art, ship path planning adopts static global planning, making it difficult to grasp the real-time situation of port traffic congestion, which makes it difficult to dynamically adjust the ship's ship path, and it is impossible to avoid congestion-related passports, resulting in poor planning results.
By obtaining the historical port throughput data of all passive ports between the starting point and end point of the ship's trip, performing serial extraction and clustering analysis, training timing prediction models, predicting port throughput and traffic status in real time, and dynamically updating the path using the associated port set, Implement path planning based on real-time port throughput.
It improves the dynamic adjustment efficiency of ship navigation path planning, can avoid congestion-related ports in real time, and ensures smooth paths.
Smart Images
Figure CN116307322B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of ship path planning, and in particular to a ship path dynamic planning method based on port throughput. Background Art
[0002] Although global path planning solves the route accessibility problem, it only considers static obstacles. Facing the unknown dynamic environment of USV navigation, global path planning technology cannot achieve safe navigation of USV. Therefore, local path planning technology is needed to correct the current route in real time so that the re-planned path can avoid dynamic and static obstacles and ensure the navigation safety of USV. According to the principle of the method, the current local path technology is mainly divided into five types: traditional algorithm, speed obstacle domain, probability sampling, dynamic window method, and reinforcement learning.
[0003] In the prior art, static global planning is usually used for the selection of ports during ship path planning, which makes it difficult to grasp the real-time situation of port congestion. The ship's sailing path planned by static global planning is difficult to dynamically adjust according to the congestion situation of the port, or after obtaining the port congestion situation, the dynamic adjustment cannot avoid the congested ports, resulting in poor ship route planning effect. Summary of the invention
[0004] The purpose of the present invention is to provide a method for dynamic planning of ship paths based on port throughput, so as to solve the technical problems that static global planning is adopted in the prior art, which makes it difficult to grasp the real-time situation of port traffic congestion, and the ship sailing path planned by static global planning is difficult to dynamically adjust according to the congestion situation of the passing port, or the dynamic adjustment after obtaining the port congestion situation cannot avoid the passing ports associated with the congestion, resulting in poor ship sailing path planning effect.
[0005] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:
[0006] A ship path dynamic planning method based on port throughput comprises the following steps:
[0007] Step S1, obtaining all the passing ports between the starting point and the end point of the ship's journey, and using the A* algorithm to plan a static ship route between the starting point and the end point of the ship's journey with the passing ports as path nodes;
[0008] Step S2, serializing and extracting the historical port throughput data of each passing port between the starting point and the end point of the ship's itinerary to obtain a historical port throughput data sequence of each passing port, and performing cluster analysis on each passing port based on the historical port throughput sequence to divide it into a plurality of associated port sets;
[0009] Step S3, using the historical port throughput data of each transit port to train a time series prediction model to obtain a throughput prediction model for each transit port;
[0010] Step S4: During the travel of the static ship route, the ship obtains the current port throughput at the next path node at the current path node, calculates the port throughput of the ship at the next path node according to the throughput prediction model, and calibrates the port traffic status at the next path node according to the port throughput of the ship at the next path node;
[0011] Step S5: Map the port traffic status at the next path node to the associated port set, determine the re-planning of the traffic port at the next path node according to the port traffic status mapping result in the associated port set, and dynamically update the static ship path at the current path node using the A* algorithm to achieve real-time dynamic planning of the ship path based on the real-time port throughput.
[0012] As a preferred solution of the present invention, the historical port throughput data of each passing port between the starting point and the end point of the ship's itinerary is serialized and extracted to obtain a historical port throughput data sequence of each passing port, including:
[0013] Extract all the time series of historical port throughput data at each passing port between the starting point and the end point of the ship's journey to obtain the total recorded time series segment, cut the total recorded time series segment using the window method to obtain a group of recorded time series segments, and measure the representativeness of the historical port throughput data series at each recorded time series segment;
[0014] The objective function for measuring the representativeness of the extracted historical port throughput data series is set by using data balance. The functional expression of the objective function is:
[0015]
[0016] Where W K is the representativeness of the historical port throughput data series at the Kth recording time segment, MK t is the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Kth recording time sequence segment at the Mth transit port, MD t is the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Dth recorded time sequence segment at the Mth passing port, n is the total number of time sequences in each recorded time sequence segment, N is the total number of recorded time sequence segments, m is the total number of passing ports, M, K, D, t are all counting variables;
[0017] A set of recorded time segments is used as a solution space to minimize the objective function, and the recorded time segments obtained by the solution are used as serialized extraction time segments;
[0018] The historical port throughput data of each transit port within the serialized extraction time series are serialized and extracted according to the time series order to obtain the historical port throughput data sequence of each transit port.
[0019] As a preferred solution of the present invention, the clustering analysis of each transit port into multiple associated port sets based on the historical port throughput sequence includes:
[0020] The passing ports corresponding to the path nodes in the static ship path are taken as cluster centers, and cluster analysis is performed on each passing port based on the cluster centers and the historical port throughput sequence to obtain multiple associated port sets.
[0021] As a preferred solution of the present invention, the throughput prediction model of each transit port is obtained by training the time series prediction model using the historical port throughput data of each transit port, including:
[0022] The historical port throughput data of each port is trained using an LSTM neural network to obtain a throughput prediction model for each port;
[0023] Among them, the input of the LSTM neural network is the historical port throughput data at the leading record time sequence of the passing port, and the output of the LSTM neural network is the historical port throughput data at the trailing record time sequence.
[0024] As a preferred solution of the present invention, the method of calculating the port throughput of the ship running to the next path node according to the throughput prediction model includes:
[0025] The current port throughput at the next path node is input into the throughput prediction model at the next path node, and the throughput prediction model at the next path node outputs a group of port throughputs at the subsequent time sequence at the next path node;
[0026] The timing of the ship running from the current path node to the next path node is calculated based on the timing of the ship at the current path node, and the timing of the ship running to the next path node is calibrated with the port throughput at a set of post-timing positions at the next path node output by the throughput prediction model to obtain the port throughput when the ship runs to the next path node.
[0027] As a preferred solution of the present invention, the port traffic status at the next path node is calibrated according to the port throughput of the ship running to the next path node, including:
[0028] The port throughput of the ship running to the next path node is input into the pre-established port traffic status evaluation model, and the port traffic status evaluation model outputs the port traffic status at the next path node;
[0029] Among them, the input of the port traffic status evaluation model is the port throughput, and the output of the port traffic status evaluation model is the port traffic status, which includes suitable for traffic and unsuitable for traffic.
[0030] As a preferred solution of the present invention, the re-planning of the pass port at the next path node is determined according to the port pass status mapping result in the associated port set, including:
[0031] When the port traffic status at the next path node is suitable for traffic, the port traffic at the next path node remains unchanged;
[0032] When the port traffic status at the next path node is unsuitable for passage, the port traffic status of the passable port at the next path node is mapped to the associated port set, and the port traffic status of all passable ports in the associated port set to which the passable port at the next path node belongs is first calibrated, and the first calibration result is unsuitable for passage;
[0033] Input the current port throughput of all the passing ports in the associated port set to which the passing port at the next path node belongs into the corresponding throughput prediction model, input the result of the throughput prediction model into the port traffic status evaluation model to obtain the port traffic status of all the passing ports in the associated port set to which the passing port at the next path node belongs, and use the port traffic status output by the port traffic status evaluation model as the second calibration result;
[0034] Combining the weights of the first calibration result and the second calibration result for all the pass ports in the associated port set to which the pass port at the next path node belongs, and using the result of the weight combination as the port passage state calibration result of all the pass ports in the associated port set to which the pass port at the next path node belongs;
[0035] The passing ports whose port traffic status is calibrated as unsuitable for traffic in all the passing ports in the associated port set to which the passing port at the next path node belongs are eliminated, and all the remaining passing ports in the associated port set to which the passing port at the next path node belongs and all the passing ports in all the associated port sets to which the passing port at non-next path nodes belongs are used as re-planning candidate ports for the passing port at the next path node.
[0036] As a preferred solution of the present invention, the method of dynamically updating the static ship path using the A* algorithm at the current path node includes:
[0037] The static ship route is re-planned using the A* algorithm from the current route node to the destination of the ship's journey, with the re-planned candidate ports as the route nodes, so as to achieve dynamic update of the static ship route.
[0038] As a preferred solution of the present invention, when the port of passage at the next path node remains unchanged, the static ship path is kept unchanged at the current path node.
[0039] As a preferred solution of the present invention, the weight in the weight combination of the first calibration result and the second calibration result is determined by using a hierarchy analysis method or an entropy weight method.
[0040] Compared with the prior art, the present invention has the following beneficial effects:
[0041] The present invention serializes and extracts the historical port throughput data of each pass port between the starting point and the end point of the ship's itinerary to obtain the historical port throughput data sequence of each pass port, and clusters and analyzes each pass port into a plurality of associated port sets based on the historical port throughput sequence, maps the port traffic status at the next path node to the associated port set, determines the re-planning of the pass port at the next path node according to the port traffic status mapping result in the associated port set, and dynamically updates the static ship route at the current path node using the A* algorithm to realize real-time dynamic planning of the ship route based on the real-time port throughput, and after obtaining the port traffic status, uses the associated port set to avoid the congested associated pass ports as a whole, thereby improving the efficiency of dynamic adjustment of the ship's route planning. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.
[0043] Figure 1 A flow chart of a method for dynamic ship path planning provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0045] In the process of ship path planning, static global planning is usually used for the selection of transit ports, which makes it difficult to grasp the real-time situation of port traffic congestion. The ship sailing path planned by static global planning is difficult to dynamically adjust according to the congestion situation of the transit port, or the dynamic adjustment after obtaining the port congestion situation cannot avoid the congested transit ports, resulting in poor ship sailing path planning effect. Therefore, the present invention provides a ship path dynamic planning method for port throughput, which can perform local dynamic adjustment on the global planned path based on the traffic status monitoring of the transit port on the basis of global planning of the ship path, so that the ship sailing path is smooth and congestion is avoided.
[0046] like Figure 1 As shown, the present invention provides a ship path dynamic planning method based on port throughput, comprising the following steps:
[0047] Step S1, obtaining all the passing ports between the starting point and the end point of the ship's journey, and using the A* algorithm to plan a static ship route between the starting point and the end point of the ship's journey with the passing ports as path nodes;
[0048] Step S2, serializing and extracting the historical port throughput data of each passing port between the starting point and the end point of the ship's itinerary to obtain a historical port throughput data sequence of each passing port, and performing cluster analysis on each passing port based on the historical port throughput sequence to divide it into a plurality of associated port sets;
[0049] Step S3, using the historical port throughput data of each transit port to train a time series prediction model to obtain a throughput prediction model for each transit port;
[0050] Step S4: During the travel of the static ship route, the ship obtains the current port throughput at the next path node at the current path node, calculates the port throughput of the ship at the next path node according to the throughput prediction model, and calibrates the port traffic status at the next path node according to the port throughput of the ship at the next path node;
[0051] Step S5: Map the port traffic status at the next path node to the associated port set, determine the re-planning of the traffic port at the next path node according to the port traffic status mapping result in the associated port set, and dynamically update the static ship path at the current path node using the A* algorithm to achieve real-time dynamic planning of the ship path based on the real-time port throughput.
[0052] The present invention realizes real-time dynamic planning of ship routes based on real-time port throughput, and after obtaining the port traffic status, uses the associated port set to comprehensively avoid the congested traffic ports, thereby improving the efficiency of dynamic adjustment of ship route planning.
[0053] Throughput is one of the indicators that characterize the congestion status of a port of passage. Constructing the throughput correlation between various ports of passage can reflect the congestion correlation between various ports of passage, and further demonstrate the correlation between the passage status of various ports of passage.
[0054] When constructing the throughput correlation of each transit port, it is necessary to select a representative sequence for correlation analysis. Therefore, the present invention avoids inaccurate correlation analysis results caused by randomly selected data sequences. By taking data balance as the optimization goal and data variance as the measurement indicator of data balance, the recording time segment with the minimum variance of the remaining recording time segments is selected. The recording time segment has the highest distribution frequency among all recording time segments, that is, the recording time segment has the highest representativeness among all recording time segments, which conforms to the throughput time series law of the transit port. Therefore, the present invention provides a method for serializing and extracting the historical port throughput data sequence of each transit port, which is as follows:
[0055] The historical port throughput data of each port between the starting point and the end point of the ship's journey are serialized and extracted to obtain the historical port throughput data sequence of each port, including:
[0056] Extract all the time series of historical port throughput data at each passing port between the starting point and the end point of the ship's journey to obtain the total recorded time series segment, cut the total recorded time series segment using the window method to obtain a group of recorded time series segments, and measure the representativeness of the historical port throughput data series at each recorded time series segment;
[0057] The objective function for measuring the representativeness of the extracted historical port throughput data series is set by using data balance. The functional expression of the objective function is:
[0058]
[0059] Where W K is the representativeness of the historical port throughput data series at the Kth recording time segment, MK t is the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Kth recording time sequence segment at the Mth transit port, MD tis the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Dth recorded time sequence segment at the Mth passing port, n is the total number of time sequences in each recorded time sequence segment, N is the total number of recorded time sequence segments, m is the total number of passing ports, M, K, D, t are all counting variables;
[0060] The objective function is minimized and solved with a set of recorded time segments as the solution space, and the recorded time segments obtained by the solution are used as the serialized extraction time segments;
[0061] The historical port throughput data of each transit port within the serialization extraction time series segment is serialized and extracted according to the time series order to obtain the historical port throughput data sequence of each transit port.
[0062] When constructing the throughput correlation of each transit port within the serialized extraction time segment, it complies with the throughput rules of each port to ensure the accuracy of the correlation analysis.
[0063] Cluster analysis of each port throughput series is performed to divide each port into multiple related port sets, including:
[0064] The passing ports corresponding to the path nodes in the static ship path are taken as cluster centers, and cluster analysis is performed on each passing port based on the cluster centers and the historical port throughput sequence to obtain multiple associated port sets.
[0065] By using cluster analysis, taking the passable ports corresponding to the path nodes in the static shipping path as the cluster centers, a set of associated ports with throughput correlation with the passable ports corresponding to the path nodes in the static shipping path can be obtained, that is, the passable ports corresponding to the path nodes in the static shipping path in each associated port set are taken as the centers, and the remaining passable ports are all associated ports with throughput correlation with the passable ports corresponding to the path nodes in the static shipping path. Furthermore, when the passable ports corresponding to the path nodes in the static shipping path are congested and unsuitable for passage, all the passable ports in the associated port set to which the passable ports corresponding to the path nodes in the static shipping path belong are congested and unsuitable for passage, thereby realizing the collective elimination of the unsuitable ports in the subsequent re-planning, avoiding the re-planned static shipping path still containing congested and unsuitable passable ports, and collectively eliminating all unsuitable ports, thereby improving the efficiency of path planning.
[0066] The present invention constructs a throughput prediction model and a port traffic status evaluation model, which can predict the port throughput (port traffic status) of the traffic port at the future time at the current time, and then predict the port throughput (port traffic status) of the traffic port corresponding to the next path node at the current traffic port corresponding to the current path node of the static ship path, and realize the pre-mastery of the port throughput (port traffic status) of the traffic port corresponding to the next path node at the current time, so that the static ship path can be adjusted in advance, and the current path node dynamically adjusts the subsequent path node, which is as follows:
[0067] The throughput prediction model of each port is obtained by training the time series prediction model using the historical port throughput data of each port, including:
[0068] The historical port throughput data of each port is trained using the LSTM neural network to obtain the throughput prediction model of each port.
[0069] Among them, the input of the LSTM neural network is the historical port throughput data at the leading record time sequence of the passing port, and the output of the LSTM neural network is the historical port throughput data at the trailing record time sequence.
[0070] The port throughput of the ship at the next path node is calculated based on the throughput prediction model, including:
[0071] The current port throughput at the next path node is input into the throughput prediction model at the next path node, and the throughput prediction model at the next path node outputs a group of port throughputs at the subsequent time sequence at the next path node;
[0072] The timing of the ship running from the current path node to the next path node is calculated based on the timing of the ship at the current path node, and the timing of the ship running to the next path node is calibrated with the port throughput at a set of post-timing positions at the next path node output by the throughput prediction model to obtain the port throughput when the ship runs to the next path node.
[0073] The port traffic status at the next path node is calibrated according to the port throughput of the ship running to the next path node, including:
[0074] The port throughput of the ship running to the next path node is input into the pre-established port traffic status evaluation model, and the port traffic status evaluation model outputs the port traffic status at the next path node;
[0075] Among them, the input of the port traffic status evaluation model is the port throughput, and the output of the port traffic status evaluation model is the port traffic status, which includes suitable for traffic and unsuitable for traffic.
[0076] The port traffic status at the acquired path node is mapped to the associated port set, the traffic status of the remaining passable ports in the associated port set to which the path node belongs is determined, and the traffic status is determined by superimposing the current throughput conditions of the remaining passable ports in the associated port set to which the path node belongs. The traffic status determined by the two situations is weighted and combined to obtain the final traffic status, which can avoid erroneous elimination in the collective elimination of all ports that are not suitable for passage, and further improve the accuracy of re-planning of the passable port at the next path node.
[0077] The replanning of the pass port at the next path node is determined according to the port pass status mapping result in the associated port set, including:
[0078] When the port traffic status at the next path node is suitable for traffic, the port traffic at the next path node remains unchanged;
[0079] When the port traffic status at the next path node is unsuitable for passage, the port traffic status of the passable port at the next path node is mapped to the associated port set, and the port traffic status of all passable ports in the associated port set to which the passable port at the next path node belongs is first calibrated, and the first calibration result is unsuitable for passage;
[0080] Input the current port throughput of all the passing ports in the associated port set to which the passing port at the next path node belongs into the corresponding throughput prediction model, input the result of the throughput prediction model into the port traffic status evaluation model to obtain the port traffic status of all the passing ports in the associated port set to which the passing port at the next path node belongs, and use the port traffic status output by the port traffic status evaluation model as the second calibration result;
[0081] Combining the weights of the first calibration result and the second calibration result for all the pass ports in the associated port set to which the pass port at the next path node belongs, and using the result of the weight combination as the port passage state calibration result of all the pass ports in the associated port set to which the pass port at the next path node belongs;
[0082] The passing ports whose port traffic status is calibrated as unsuitable for traffic in all the passing ports in the associated port set to which the passing port at the next path node belongs are eliminated, and all the remaining passing ports in the associated port set to which the passing port at the next path node belongs and all the passing ports in all the associated port sets to which the passing port at non-next path nodes belongs are used as re-planning candidate ports for the passing port at the next path node.
[0083] The static ship path is dynamically updated using the A* algorithm at the current path node, including:
[0084] The static ship route is re-planned using the A* algorithm from the current route node to the destination of the ship's journey, with the re-planned candidate ports as the route nodes, so as to achieve dynamic update of the static ship route.
[0085] When the port of passage at the next path node remains unchanged, the static ship path is kept unchanged at the current path node.
[0086] The weight in the weight combination of the first calibration result and the second calibration result is determined by using a hierarchy analysis method or an entropy weight method.
[0087] The present invention serializes and extracts the historical port throughput data of each pass port between the starting point and the end point of the ship's itinerary to obtain the historical port throughput data sequence of each pass port, and clusters and analyzes each pass port into a plurality of associated port sets based on the historical port throughput sequence, maps the port traffic status at the next path node to the associated port set, determines the re-planning of the pass port at the next path node according to the port traffic status mapping result in the associated port set, and dynamically updates the static ship route at the current path node using the A* algorithm to realize real-time dynamic planning of the ship route based on the real-time port throughput, and after obtaining the port traffic status, uses the associated port set to avoid the congested associated pass ports as a whole, thereby improving the efficiency of dynamic adjustment of the ship's route planning.
[0088] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.
Claims
1. A ship path dynamic planning method based on port throughput, characterized in that: The following steps are involved: Step S1, obtaining all the passing ports between the starting point and the end point of the ship's journey, and using the A* algorithm to plan a static ship route between the starting point and the end point of the ship's journey with the passing ports as path nodes; Step S2, serializing and extracting the historical port throughput data of each passing port between the starting point and the end point of the ship's itinerary to obtain a historical port throughput data sequence of each passing port, and performing cluster analysis on each passing port based on the historical port throughput sequence to divide it into a plurality of associated port sets; Step S3, using the historical port throughput data of each transit port to train a time series prediction model to obtain a throughput prediction model for each transit port; Step S4: During the travel of the static ship route, the ship obtains the current port throughput at the next path node at the current path node, calculates the port throughput of the ship at the next path node according to the throughput prediction model, and calibrates the port traffic status at the next path node according to the port throughput of the ship at the next path node; Step S5: Map the port traffic status at the next path node to the associated port set, determine the re-planning of the traffic port at the next path node according to the port traffic status mapping result in the associated port set, and dynamically update the static ship path at the current path node using the A* algorithm to achieve real-time dynamic planning of the ship path based on the real-time port throughput; The historical port throughput data of each passing port between the starting point and the end point of the ship's itinerary are serialized and extracted to obtain a historical port throughput data sequence of each passing port, including: Extract all the time series of historical port throughput data at each passing port between the starting point and the end point of the ship's journey to obtain the total recorded time series segment, cut the total recorded time series segment using the window method to obtain a group of recorded time series segments, and measure the representativeness of the historical port throughput data series at each recorded time series segment; The objective function for measuring the representativeness of the extracted historical port throughput data series is set by using data balance. The functional expression of the objective function is: Where W K is the representativeness of the historical port throughput data series at the Kth recording time segment, MK t is the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Kth recording time sequence segment at the Mth transit port, MD t is the historical port throughput data at the tth time sequence in the historical port throughput data sequence at the Dth recorded time sequence segment at the Mth passing port, n is the total number of time sequences in each recorded time sequence segment, N is the total number of recorded time sequence segments, m is the total number of passing ports, M, K, D, t are all counting variables; A set of recorded time segments is used as a solution space to minimize the objective function, and the recorded time segments obtained by the solution are used as serialized extraction time segments; The historical port throughput data of each transit port within the serialized extraction time series are serialized and extracted according to the time series order to obtain the historical port throughput data sequence of each transit port.
2. A ship path dynamic planning method based on port throughput according to claim 1, characterized in that: The cluster analysis of each transit port based on the historical port throughput sequence is divided into multiple associated port sets, including: The passing ports corresponding to the path nodes in the static ship path are taken as cluster centers, and cluster analysis is performed on each passing port based on the cluster centers and the historical port throughput sequence to obtain multiple associated port sets.
3. A ship path dynamic planning method based on port throughput according to claim 2, characterized in that: The throughput prediction model of each port is obtained by training the time series prediction model using the historical port throughput data of each port, including: The historical port throughput data of each port is trained using an LSTM neural network to obtain a throughput prediction model for each port; Among them, the input of the LSTM neural network is the historical port throughput data at the leading record time sequence of the passing port, and the output of the LSTM neural network is the historical port throughput data at the trailing record time sequence.
4. A ship path dynamic planning method based on port throughput according to claim 3, characterized in that: The method of calculating the port throughput of the ship at the next path node according to the throughput prediction model includes: The current port throughput at the next path node is input into the throughput prediction model at the next path node, and the throughput prediction model at the next path node outputs a group of port throughputs at the subsequent time sequence at the next path node; The timing of the ship running from the current path node to the next path node is calculated based on the timing of the ship at the current path node, and the timing of the ship running to the next path node is calibrated with the port throughput at a set of post-timing positions at the next path node output by the throughput prediction model to obtain the port throughput when the ship runs to the next path node.
5. A ship path dynamic planning method based on port throughput according to claim 4, characterized in that: The method of calibrating the port traffic status at the next path node according to the port throughput of the ship running to the next path node includes: The port throughput of the ship running to the next path node is input into the pre-established port traffic status evaluation model, and the port traffic status evaluation model outputs the port traffic status at the next path node; Among them, the input of the port traffic status evaluation model is the port throughput, and the output of the port traffic status evaluation model is the port traffic status, which includes suitable for traffic and unsuitable for traffic.
6. A ship path dynamic planning method based on port throughput according to claim 5, characterized in that: The re-planning of the pass port at the next path node is determined according to the port pass status mapping result in the associated port set, including: When the port traffic status at the next path node is suitable for traffic, the traffic port at the next path node remains unchanged; When the port traffic status at the next path node is unsuitable for passage, the port traffic status of the passable port at the next path node is mapped to the associated port set, and the port traffic status of all passable ports in the associated port set to which the passable port at the next path node belongs is first calibrated, and the first calibration result is unsuitable for passage; Input the current port throughput of all the passing ports in the associated port set to which the passing port at the next path node belongs into the corresponding throughput prediction model, input the result of the throughput prediction model into the port traffic status evaluation model to obtain the port traffic status of all the passing ports in the associated port set to which the passing port at the next path node belongs, and use the port traffic status output by the port traffic status evaluation model as the second calibration result; Combining the weights of the first calibration result and the second calibration result for all the pass ports in the associated port set to which the pass port at the next path node belongs, and using the result of the weight combination as the port passage state calibration result of all the pass ports in the associated port set to which the pass port at the next path node belongs; The passing ports whose port traffic status is calibrated as unsuitable for traffic in all the passing ports in the associated port set to which the passing port at the next path node belongs are eliminated, and all the remaining passing ports in the associated port set to which the passing port at the next path node belongs and all the passing ports in all the associated port sets to which the passing port at non-next path nodes belongs are used as re-planning candidate ports for the passing port at the next path node.
7. A ship path dynamic planning method based on port throughput according to claim 6, characterized in that: The method of dynamically updating the static ship path using the A* algorithm at the current path node includes: The static ship route is re-planned using the A* algorithm from the current route node to the destination of the ship's journey, with the re-planned candidate ports as the route nodes, so as to achieve dynamic update of the static ship route.
8. A ship path dynamic planning method based on port throughput according to claim 6, characterized in that: When the port of passage at the next path node remains unchanged, the static ship path is kept unchanged at the current path node.
9. A ship path dynamic planning method based on port throughput according to claim 6, characterized in that: The weight in the weight combination of the first calibration result and the second calibration result is determined by using a hierarchy analysis method or an entropy weight method.